Gardens’ Bulletin Singapore 63(1 & 2): 329–339. 2011 329 The floristic position of Java Peter C. van Welzen1 and Niels Raes Netherlands Centre for Biodiversity – Naturalis, Section National Herbarium of The Netherlands, Leiden University, PO Box 9514, 2300 RA Leiden, The Netherlands [email protected] (corresponding author) ABSTRACT. Floristically, Java was always considered to be west of Wallace’s line together with the Malay Peninsula, Sumatra, Borneo, the Philippines and Sulawesi. Recently, statistical analyses of rough geographic data per species (presence or absence on islands or island groups) showed that Java is part of the central Wallacean area in Malesia (together with also the Philippines, Sulawesi, the Lesser Sunda Islands, and the Moluccas) rather than of Sundaland (restricted to the Malay Peninsula, Sumatra, and Borneo). More precise distribution maps for Java with collecting localities show that most species are widespread over Java or show a (more) western distribution; few species show an eastern distribution. The distributions show strong correlations with altitude (mountain species) and with precipitation (roughly wet in the west, dry in the east). The expectation was to find mainly species with a drought preference (Wallacean). However, most species show a preference for a wet distribution, which is related to a Sunda distribution. The fact that the statistical tests used for the first database show a Wallacean connection for Java probably is the result of the relative values these test use instead of absolute numbers, e.g., the resemblance between, especially, the flora of the Lesser Sunda Islands with Java is very high. Keywords. Flora Malesiana, floristics, Java, Malesia Introduction In 1859 Wallacea introduced his famous zoological boundary, Wallace’s line (Huxley 1868), that divided the Malay Archipelago (or Malesia: Steenis 1950; Raes & Welzen 2009) into an eastern and western part. Wallace’s line runs east of the Philippines, then either west (Wallace 1859, 1863–1876) or east (Wallace 1860, 1910) of Sulawesi (also known as Celebes), and ends between Bali and Lombok in the Lesser Sunda Islands. Wallace discussed the position of Sulawesi in his book ‘Island life’ (Wallace 1880), in which he calls Sulawesi an ‘anomalous island’ with no continental connections as Sulawesi lacks Sundaic groups and contains (old) endemic and Australasian species. A more complete historical overview is presented in Simpson (1977) and George (1981), who both show that a number of variants of Wallace’s line have been proposed based on the study of different groups of organisms (Fig. 1). The area encompassed by these lines is often called Wallacea, a term coined by Dickerson (1928), for an area already delimited by Wallace in 1863. The areas to the west (Malay Peninsula, Sumatra, Java, Borneo) and to the east (New Guinea) are referred to as the Sunda Shelf and the Sahul Shelf, respectively. 330 Gard. Bull. Singapore 63(1 & 2) 2011 Fig. 1. Interpretations of Wallace’s lines around Sulawesi and various alternatives are depicted in shades of blue. Red lines show the borders between the Sunda Shelf, Wallacea and the Sahul Shelf resulting from analyses discussed in the text. Almost only zoological data were used to distinguish the lines (George 1981), and unsurprisingly, most botanist did not use the various boundaries. Steenis (1950) used distributions of plant genera when he contemplated the limits of the Malesian region, and based on the same data Wallace’s line did not appear to be a distinct boundary in plant distributions. Welzen et al. (2005) used a limited floristic species database to show that all lines do form distinctive boundaries in plant distributions, each line stopping at least twice as many taxa as passing. Welzen & Slik (2009) even indicated which families are mainly responsible for the distribution patterns in Malesia. Only 20 families determine the patterns. Dipterocarpaceae, Fagaceae, and Nepenthaceae have their centre of diversity on the Sunda Shelf; Ericaceae, Monimiaceae, and Sapindaceae are typical for the Sahul Shelf; Araliaceae, Boraginaceae, Convolvulaceae, Cyperaceae, Dioscoreaceae, Lamiaceae, Loranthaceae, Mimosaceae, and Moraceae are predominantly Wallacean, and Burseraceae, Caesalpiniaceae, Flacourtiaceae, Meliaceae, and Myristicaceae show no distinct centre. Recently, Welzen et al. (2011) analysed the distribution data per island group of all species published in Flora Malesiana series 1 (Angiosperms) and the Malesian orchids in Orchids Monograph (for island groups see Fig. 1). Various phenetic techniques were used: Principal Components Analysis, Fig. 2A; Non-metric Multidimensional Scaling analysis, Fig. 2B; a cluster analysis, Unweighted Pair Group Method with Arithmetic mean, Fig. 2C; and Kroeber’s coefficient, the mean Floristic position of Java 331 floral similarity between pairs of areas, Fig. 2D. All results show that Java is not west of Wallace’s line, but east of it (Fig. 2). The tests were repeated for 100 randomly drawn matrices with equal contributions of all areas (500 species per area), which should nullify the effect of different island sizes (large islands like Borneo and New Guinea harbour far more (endemic) species than other areas and are then automatically separated from the rest in the analyses). Also, the analyses of the 100 matrices showed the same result for Java, not a part of the Sunda Shelf areas, but part of Wallacea. Thus, the borders between the three areas, Sunda Land/Shelf, Wallacea and Sahul Land/ Shelf, have to be redrawn a bit (red lines in Fig. 1). The Kroeber’s Coefficient (Fig. 2D) nicely shows that in fact the botanical relationship between Java and Wallacea just wins in the other three analyses (PCA, NMS, UPGMA) from a Sunda relationship, because Java has a high floral mean resemblance (Fig. 3) with on the one hand Sumatra and the Malay Peninsula (Sunda Land) and on the other hand a slightly higher mean resemblance with the Lesser Sunda Islands and Sulawesi (Wallacea). The floral resemblance with Borneo (Sunda) and the Philippines and Moluccas (Wallacea) is somewhat less. Welzen et al. (2011) Fig. 2. The results of various phenetic tests on a database with presence/absence data per island group for indigenous species revised in Flora Malesiana and Orchid Monographs. All tests place Java together with Wallacean areas. A. Principle Components Analysis (PCA). B. Non- metric Multidimensional Scaling analysis (NMS). C. Unweighted Pair Group Method with Arithmetic mean (UPGMA). D. Kroeber’s coefficients with a mean floral similarity between pairs of areas indicated by various thicknesses of the lines connecting the pairs. 332 Gard. Bull. Singapore 63(1 & 2) 2011 Fig. 3. Kroeber’s coefficients between Java and the Sunda and Wallacean areas showing a high resemblance, on the one hand, with Sumatra and the Malay Peninsula on the Sunda Shelf; and, on the other hand, with the Lesser Sunda Islands and Sulawesi within Wallacea. Blue lines depict the borders of Wallacea. discuss that the floral resemblances can largely be explained by the supposed savannah corridors running from the Malay Peninsula along and over Sumatra to Java during glacial periods. Via this corridor species that prefer a yearly dry period could disperse and are now still found in areas with a yearly dry monsoon. The purpose of this paper is threefold: - establish why plant distributions more strongly point to a Wallacean relationship for Java; - check which kind of phenetic distribution patterns exist witin Java; and - explain these. Materials and methods In previous studies (Welzen et al. 2005, Welzen & Slik 2009, Welzen et al. 2011), a database was used whereby the presence and absence of indigenous species revised in Flora Malesiana series 1 and Orchid Monographs was noted for the Malay Peninsula, Sumatra, Java, Borneo, the Philippines, Sulawesi, Lesser Sunda Islands, Moluccas, and New Guinea. The areas used necessitated two alterations for two lines. The Merrill- Dickerson variant of the Huxley line runs officially between Palawan and the rest of the Philippines. This line is now considered to run between Palawan and Borneo (Fig. 1). Also, Wallace’s line ends between Bali and Lombok in the Lesser Sunda Islands; this is redrawn between Java and Bali (Fig. 1). These redrawn lines are used in this study. For this study, a database was created with collecting localities of species that are represented by digitised herbarium specimens from Java. Up to now, the database Floristic position of Java 333 contains 97 families and 447 genera. The families starting with A or B are fully digitised, plus a part of C (26 families), some of the other families have also been digitised (e.g., Euphorbiaceae, Rhamnaceae, Sapindaceae, Vitaceae), but most other families are only partly digitised. The localities were georeferenced with the websites http://earth-info. nga.mil/gns/html/index.html and http://www.fallingrain.com/world/. Only the species with five or more different localities were included in the analysis, fewer localities were considered too incomplete to infer a distribution pattern. In total 808 species were sorted visually into different distribution patterns. This appeared to be rather straightforward. Four patterns were discriminated: 1) widespread over the island; 2) a western distribution (subdivided into 2a purely west, 2b west plus a few dots in central Java, and 2c west up to central Java; these three were not always distinctive); 3) a (central to) eastern distribution; and 4) a west-and-east-only distribution. All distributions were divided into low and high (above 1000 m) altitude. The dot maps were produced with MapInfo Professional 7.0 (© MapInfo Corporation), the altitudinal and precipitation maps were made with Manifold GIS (Manifold.net), for which the datasets were obtained from the WorldClim 30 arc- second dataset (www.worldclim.org). Results Table 1 shows how many species were found per pattern. A large part, 49%, of the sampled plant species is widespread over Java (Fig. 4A). This group was not used in further analyses as they did not convey any information concerning a possible split of Java into the Sunda or Wallacean realms.